more mxe testing

[Mike Miller]

I've built the stable and development versions of Octave with mxe and
tested both under wine. Here's a quick list of the issues I've noticed
with both. Some of these may be due to wine, maybe someone with a real
windows can tell which of these are real and which are wine artifacts.

* Qt timer errors, gui crash (I think discussed in another thread)

* gnuplot doesn't come up at all, octave just hangs (gnuplot does work
in wine when called on its own)

* first fltk plot comes up initially all black until a mouse drag event

* a handful of numerical precision errors, including things like

octave:1> format long
octave:2> 10^-10
ans = 1.00000000000000e-010
octave:3> 1e-10
ans = 1.00000014901162e-010

* slash options are treated as paths ("dir /d" and "sort /r" both return
a "File not found" error message)

* umask has no effect on file permissions (failure in test_system.m)

The file permissions thing makes sense, this test should probably just
be bypassed on windows.

Is a different command interpreter used for the system() function
depending on whether something is built with cygwin or mingw or native?
How does one know whether forward slashes passed to system() will be
interpreted as paths or command options?

Also, the attached changeset adds texinfo to the dependencies for the
stable Octave build.

-- 
mike

mxe-octave-texinfo.diff
Description: Text Data